The Social Citizen

Peer Networks and Political Behavior

Betsy Sinclair

• Publisher: University of Chicago Press
• Year: 2012
• ISBN:
  9780226922836
• Topic: Politica e relazioni internazionali
• Subtopic: Politica

Human beings are social animals. Yet despite vast amounts of research into political decision making, very little attention has been devoted to its social dimensions. In political science, social relationships are generally thought of as mere sources of information, rather than active influences on one's political decisions. Drawing upon data from settings as diverse as South Los Angeles and Chicago's wealthy North Shore, Betsy Sinclair shows that social networks do not merely inform citizen's behavior, they can—and do—have the power to change it. From the decision to donate money to a campaign or vote for a particular candidate to declaring oneself a Democrat or Republican, basic political acts are surprisingly subject to social pressures. When members of a social network express a particular political opinion or belief, Sinclair shows, others notice and conform, particularly if their conformity is likely to be highly visible. We are not just social animals, but social citizens whose political choices are significantly shaped by peer influence. The Social Citizen has important implications for our concept of democratic participation and will force political scientists to revise their notion of voters as socially isolated decision makers.

1

Introduction: Social Pressure and Participatory Democracy

What motivates a nonvoter to participate in politics? Why do some individuals donate so much money to political candidates and organizations? What motivates a registered Democrat to vote for a Republican candidate? Why would a college student choose to register for the Republican Party when the rest of her family are registered Democratic? This book addresses these questions with the larger goal of identifying what may be the most critical determinant of political behavior. Specifically, in contrast to common portrayals of political actors as atomistic, citizens’ political actions depend largely on their social interactions. Making sense of citizens’ political decisions requires a sustained understanding of their social networks.

Politics are incredibly contagious in social networks. Shared political behaviors of an individual’s social network affect both participation and political choices. When friends and family talk about politics, they are referring to tightly held personal norms of civic behavior, and in close personal relations it is difficult to disagree about such beliefs. As Michael MacKuen (1990, 94) notes, “Politics probably incorporates more of a sense of morality than many other topics of conversation.” Perhaps, then, people are cautious to whom they talk about politics (Eliasoph 1998). Yet voters do talk to their families and friends about politics regularly. In the 2004 American National Election Studies (ANES), 80 percent of the respondents reported having conversations with their families and friends about politics.¹

What are these friends talking about? Do these social interactions matter? There are good reasons to think not. Indeed, these friends are unlikely to be sharing information. One of the most notable features of democratic citizens is their lack of political knowledge. An extensive literature documents the lack of political interest, information, or sophistication among most citizens. Voters may not know the number of senators from their state, for example, and, less surprisingly, may have difficulty locating themselves and the candidates on an ideological spectrum, remembering the names of candidates, or knowing where any particular candidate stands on a broad set of issues (Almond and Verba 1963; Berelson, Lazarsfeld, and McPhee 1954; Campbell et al. 1960; Converse 1964; Kinder and Sears 1985). Voters are often reported to be poorly informed about the candidates’ positions (Campbell et al. 1960; Palfrey and Poole 1987; Alvarez 1998). Given this ostensible disengagement, is it plausible that social interactions play a pivotal role in shaping political attitudes and behaviors? Political conversations need not be particularly informative to have an effect: it need not be information transmission that drives the impact of social groups.

Instead, individuals are likely to be talking about politics in the context of constructing a social and political identity. There are two possibilities whereby individuals are influenced by their social environment via explicitly external channels.² First, as discussed above, some individuals are persuaded by information, shared between their social ties, to change their political behavior. In this instance, political persuasion occurs as a consequence of explicit social communication. Persuasion, some argue, occurs as a consequence of deliberation or exposure to particular rhetoric. The second channel is that of conformity to a social norm for purely social purposes. In this case, individuals mirror the political norms of those within their social network who have strong normative preferences to particular political behaviors, like voting or donating to campaigns. This last channel can help to simply maintain a pattern of behavior. This is distinct from compliance, where an individual adopts a particular behavior based upon the coercive power of the social network. In this last channel, individuals adopt norms they believe are correct as well as socially desirable for their social ties. It is not necessary for one individual to directly request compliance from another individual, although personal requests are possible. It is sufficient for one individual to observe the behavior of another through regular social interactions: this kind of interaction facilitates social learning about political norms.

This chapter proceeds by characterizing an individual’s political network. Political networks may influence an individual’s politics via two primary mechanisms—information and social pressure—and the section that follows distinguishes these two mechanisms and the kinds of empirical patterns that emerge from each one. The remainder of the book is then outlined in the following sections, with particular attention to its methodological contributions and the diversity of approaches for identifying political networks.

A Political Network

Citizens’ social networks describe the collection of individuals tied to them by social connections, such as friendships, family relationships, or work colleagues. Each individual’s social network forms by some combination of choice, where the individual purposefully selects network members, and chance, where the individual randomly forms relationships with others. A political network consists of a subset of these social peers with whom an individual discusses politics, elections, or government. Political networks overlay preexisting social networks; they are not necessarily geographically based but are founded on strong social relationships, likely dominated by primary group members with whom the individual communicates about external social norms.

Throughout this book, different techniques are presented to establish an individual’s political network. Unless the individual has been solicited to identify specific individuals with whom she has political interactions, however, the collection of relationships presented encompass her larger social network. Insight into what the larger social network structure looks like can be drawn from sociology and microeconomics, where a significant amount of work has been done to describe the formulation of social networks. In this literature, individuals are described as nodes in a graph where links (either directed or not) represent connections between people (friendships, coauthors, or simply people who know each other). Examples of social networks abound: a favorite game for movie lovers, Six Degrees of Kevin Bacon; the social networking website Facebook; and Milgram’s (1967) classic study of “small worlds,” in which he examined how many times a letter was handled before it was received by the intended recipient. The shape of the social network is determined by where links are formed, their distribution, and their density. Political networks are subsets of social networks that are composed of links formed both randomly and by individuals searching for other individuals.³ Individuals tend to have political networks with low density—that is, in very few cases are the political discussion partners of one individual socially connected to each other—and to be asymmetric: an individual may identify someone as a political discussant, but she may not necessarily reciprocate (Huckfeldt and Sprague 1995; Huckfeldt, Johnson, and Sprague 2004).

Numerous empirical studies have identified the presence of political networks, and a number of national probability sample surveys have incorporated items designed to document their presence.⁴ The 2000 ANES social network battery offered insight into the interaction of disagreement and social ties (Huckfeldt, Johnson, and Sprague 2004). This study clarified that political networks tend not to be very large. In the 2000 ANES, only 18 percent of the survey respondents could name four individuals with whom they discuss politics. Over 40 percent of the discussants were relatives of the respondent, 25 percent were coworkers, 10 percent attended the same church, and 20 percent were neighbors. The average respondent reported speaking to each discussant a couple of times per week.

Huckfeldt and Sprague (1995) asked respondents specifically about their political contacts and then interviewed the contacts to get a more complete understanding of an individual’s network. They examined both the 1984 and 1996 presidential elections looking for the existence and effects of political networks. In examining the 1984 presidential election they focused on the effects of an individual’s preferences (and the distribution of preferences) on the choice of “political discussion partners.”⁵ Their survey asked respondents whom they cast their votes for and how they believed their discussion partners cast their votes. Their results indicate that individuals are more likely to have conversations with people who agree with them but that there is some amount of political heterogeneity. They found that two-thirds of Reagan voters had discussants who reported voting for Reagan and that 57 percent of Mondale voters had discussants who reported voting for Mondale.⁶ Political networks are generally homogeneous in terms of socioeconomic and demographic characteristics but are not completely characterized by political homogeneity. Political networks can form as an unintended consequence of social groups or for actual political purposes.

The observed correlation between the survey respondents’ and their discussants’ choices in the Huckfeldt and Sprague Reagan-Mondale results is likely to be present within most political networks. This correlation could be attributable to a history of political conversations with some of their network members in each election cycle, for example, or could be attributable to the selection of politically similar network members. This type of correlation makes it particularly difficult to identify network effects. This book employs a range of empirical strategies to circumvent the identification problems generated by this kind of correlation, such as taking advantage of changing preferences over time or intervening with a randomized field experiment. These empirical strategies are discussed further in the chapters that follow.

Why is there such a desire to mirror the political norms of the social network? Social scientists have long understood that most individuals have a fundamental need to belong to and affiliate with groups (Baumeister and Leary 1995).⁷ Yet it has never been theorized that the individual conforms with the group on all issues; rather, individuals are selectively pressured on those issues that are essential to the stability of the group (Steiner 1954; Walsh 2004). Disagreement with respect to social-political norms can undermine the stability of the group. Individuals are likely to have their own preferences about politics, but as elections and politics become socially visible during an election campaign, individuals feel pressure to conform to the norms of their social networks (Berelson, Lazarsfeld, and McPhee 1954; Huckfeldt and Sprague 1995). In the context of casual conversations, individuals are likely to become aware of political differences between themselves and other members of their social network.⁸ To this extent, then, there is a range of topics for which social conformity is likely to exist. Political behavior, particularly related to social norms, is likely to fall into this category periodically, particularly preceding an election. That is, groups succeed at collectively establishing political norms when politics become a salient component of the group conversation (Merei 1952; Katz and Lazarsfeld 1955; Walsh 2004). This happens as a consequence of the desire to establish a common social identity (Conover et al. 2002). Individuals face pressures to conform (Festinger 1954), and responding to these pressures leads to shared political behavior. An individual will balance her personal preferences and choices against those of her social context, and most individuals prefer to agree with their social network than to defend their individual politics. Thus individuals succumb to social pressure and adopt the political preferences and behaviors of their social network.

That individuals prefer to belong to groups with homogeneous preferences is clear from their selection of a social network. These choices are governed not by political preferences, however, but by social preferences. Individuals choose their social networks based on shared characteristics such as socioeconomic and demographic characteristics. Political characteristics are likely to be correlated to these variables and thus are also likely to be shared (Lazar et al. 2008). Lazarsfeld, Berelson, and Gaudet (1948, 137) write, “Most people interacted during the campaign with others with shared social characteristics, shared attitudes and, thus, shared political pre-dispositions.” Yet individuals primarily form social relationships based on shared nonpolitical characteristics. Weatherford (1982, 129) finds that the variables that affect the degree of social interaction between local residents “do not contribute to network politicization.”⁹ Even groups in which politics are regularly discussed are not formed based on shared political preferences (Walsh 2004).

The phenomenon of individuals preferring groups with homogeneous preferences is referred to as homophily: the tendency of an individual to associate with similar others (Lazarsfeld and Merton 1954; Coleman 1958; McPherson, Smith-Lovin, and Cook 2001). Homophily is most likely to occur among those of the same race and ethnicity, age, religion, education, occupation, and gender, roughly in that order (McPherson, Smith-Lovin, and Cook 2001). Social relationships, despite homophily with respect to socioeconomic and demographic characteristics, are unlikely to be characterized by complete political homogeneity, as political issues do not dominate interpersonal conversation or interest. Interest in politics is mainly focused on election cycles, and there is consistent evidence that individuals do disagree about their politics in the course of elections (Huckfeldt and Sprague 1995; Huckfeldt et al. 1995; Huckfeldt, Sprague, and Levine 2000; Huckfeldt, Johnson, and Sprague 2004; Ikeda and Huckfeldt 2001; Schmitt-Beck 2003). Social networks are established, then, both by choice and by circumstance (Jackson and Rogers 2007; Mollenhorst, Volker, and Flap 2008), and relationships change frequently. Approximately half of all close relationships change every seven years. Assuming the presence of new ties, upon which no political pressure has been exerted outside of an election cycle, there will be disagreement within the group regarding politics during the course of an election (Mollenhorst, Volker, and Flap 2007, 2008).

To summarize, a basic theory of human sociability explains the instances when we make collective choices or engage in collective behavior. Individuals choose homogeneous social networks. Particularly during election cycles or at other times when politics is likely to be salient in casual conversations, a political network overlays each individual’s social connections. The political network is not homogeneous because it is not formed based on political preferences but rather on other social preferences. Citizens’ political attitudes and behavior depend critically on social groups. These groups form for nonpolitical purposes, but the norms of group behavior, at critical times, lead to what otherwise appear to be surprising political actions, such as when a registered Democrat votes for a Republican candidate. Yet social networks will only influence individuals if network members give voice to a particular political norm, sincerely adhering to the political norm in an observable way. In this sense, human sociability shapes democratic outcomes. Put simply, individuals do not want to disappoint their friends and family, and this is how politics are contagious. Not only is human sociability important for understanding the origins of political behaviors, but it also raises serious questions about how democracies work, since the contagion process may not stem from political preferences but from social pressures resulting from the nonpolitical dynamics of network formation.

Information or Social Pressure

While some of the existing literature has documented the influence of political networks, they are seldom seen as a dominant explanation for political behavior, in large part because of the difficulty of identifying the mechanism that drives the observed correlations in behavior between network members. Linking the existing research on the influence of social networks to new research on four political behaviors (voting, donating, choosing a candidate, and choosing a party identification) reveals patterns that attribute a component of political behavior to social pressure exerted by a political network. Individuals are embedded in particular political networks, and their networks are a component of their decision calculus. The network preferences determine, to some extent, individual participation and choices. The type of network—whether it is one in which individuals regularly interact, for example, or whether the network is primary or secondary—determines the amount of network influence. Particular behaviors are also more or less likely to be exposed to network effects.

There are two principal theories regarding how individuals might adopt the behaviors of their political network. First, individuals may rely on their social networks for information, in particular from trusted sources. Social networks may provide arguments to use for deliberation, may provide information about candidate platforms or places where donations are needed, or may simply relay how and when to cast a ballot. Information has typically been considered contagious across political networks; this is the most common explanation for the presence of social network effects (Sokhey and McClurg 2008; Djupe and Sokhey 2011). Second, individuals may face social pressure from their political networks. Individuals rely on their social networks to maintain their social identity and may base their preferences on social comparison with their network. In this case, the network exerts social pressure on each member, particularly if a majority of network members have homogeneous preferences regarding a political or social norm. There are strong arguments underpinning each of these theories, and each theory generates a distinct set of hypotheses regarding a set of empirical regularities. For example, if social pressure dominates a citizen’s experience of democracy, then across behaviors there should be a network effect. The effect should be determined more by intimate network ties and less by a broader social context or other indirect influences. This effect should be larger for relationships in which there is a tighter social connection. In contrast, network influence generated by shared information should not vary based on the social proximity of the relationship.

Information. Suppose that a voter determines which candidate to support in an upcoming election. How could information from her political network influence her choice? Voters’ information about candidates comes from many sources before an election, some trustworthy and some not. Voters know that campaigns attempt to influence voter opinions by producing biased information, but they also know that more objective sources exist. Not all of this information is free, however. While campaigns may spend millions of dollars to bring information into voters’ living rooms, voters must actively seek out objective sources. While voters may realize that they receive heavily biased information from the campaign, they may not have the interest or time to investigate unbiased sources and procure information. There is, however, a compromise between spending time and effort reading about candidates and simply viewing campaign advertising. It is possible to rely on a trusted source, specifically someone who has procured enough information to make an informed decision and will not relay heavily biased information (Downs 1957). If there is no single individual who is willing to pay the cost to procure all the information about the candidates, it is also possible for individuals to aggregate information (McKelvey and Ordeshook 1985a, 1985b, 1990; Converse 1962). This aggregation process occurs completely naturally through friendships (Huckfeldt, Johnson, and Sprague 2004).

If political information can be aggregated through a network of friends, then no single individual need spend much time researching candidate positions. A large literature exists about how voters take cues from sources around them to minimize the costs of determining which candidate to supp...